Video-On-Demand Recommendations Based on Previously Viewed Television Programs

1. A method, comprising: maintaining television data that identifies broadcast television programs that have been previously viewed by one or more of a plurality of television viewers; maintaining VOD data that identifies videos-on-demand that have been purchased by one or more of the plurality of television viewers; associating, by a headend, a particular one of the broadcast television programs with a particular one of the videos-on-demand if the television data and the VOD data indicates that a first television viewer watched the particular broadcast television program and purchased the particular video-on-demand, the associating including assigning a weight to the particular broadcast television program that is associated with the particular video-on-demand, wherein the weight describes a strength of association between the particular broadcast television program and the particular video-on-demand and represents a conditional probability that a viewer will purchase the particular video-on-demand based on a factor that the viewer has already watched the particular broadcast television program and a probability that any viewer will purchase the particular video-on-demand; and recommending, by the headend, the particular video-on-demand to a second television viewer at a client device if the second television viewer has watched the particular broadcast television program.

2. The method as recited in claim 1 , wherein the television data comprises: a client device identifier; and a television program identifier.

3. The method as recited in claim 2 , wherein the client device identifier comprises an obfuscated form of an identifier that uniquely identifies a particular client device.

4. The method as recited in claim 1 , wherein the maintaining television data comprises maintaining data that identifies broadcast television programs that have been displayed using a particular client device for, at a minimum, a particular period of time.

5. The method as recited in claim 4 , wherein the particular period of time comprises three minutes.

6. The method as recited in claim 1 , wherein the VOD data comprises: a client device identifier; and a VOD identifier.

7. The method as recited in claim 6 , wherein the client device identifier comprises an obfuscated form of an identifier that uniquely identifies a particular client device.

8. The method as recited in claim 1 , wherein the recommending comprises verifying that the second television viewer has not purchased the particular video-on-demand.

9. The method as recited in claim 1 , wherein the recommending comprises verifying that the particular video-on-demand is currently available to be purchased or viewed.

10. The method as recited in claim 1 , wherein the weight represents a percentage of viewers who purchased or watched the particular video-on-demand and who also watched the particular broadcast television program.

11. The method as recited in claim 1 , wherein the associating comprises associating a broadcast television program identifier with a VOD identifier.

12. The method as recited in claim 11 , wherein the associating further comprises associating a weight with the broadcast television program identifier and the VOD identifier.

13. The method as recited in claim 1 , wherein the recommending comprises applying a decision tree algorithm to the VOD data and the television data to create a decision tree for each of a plurality of VOD titles using the TV data to generate tree splitting criteria.

14. The method as recited in claim 13 , wherein: the decision tree for each of the plurality of VOD titles comprises a probabilistic classification tree; and the tree splitting criteria comprises a Bayesian score.

15. A computer-readable storage device having computer-readable instructions stored thereon that, when executed by a processor, implements the method as recited in claim 1 .

16. The method as recited in claim 1 , wherein the weight is based at least in part on a frequency with which the particular broadcast television program is watched by the first television viewer.

17. A method, comprising: identifying a video-on-demand that has been purchased by a particular viewer; associating, by a headend, with the video-on-demand, a broadcast television program that has been watched by the particular viewer, the associating comprising: creating an association between a television program identifier that is associated with the broadcast television program and a VOD identifier that is associated with the video-on-demand; assigning a weight to the association, wherein the weight describes a strength of association between the television program identifier and the VOD identifier and the weight represents a conditional probability that a viewer will purchase the video-on-demand based on a factor that the viewer has already watched the particular broadcast television program and a probability that any viewer will purchase the particular video-on-demand; and recommending, from the headend to a client device, the video-on-demand to another viewer who has watched the broadcast television program.

18. The method as recited in claim 17 wherein the identifying comprises: maintaining a client device ID that is associated with the viewer; and maintaining a VOD ID that is associated with the video-on-demand.

19. The method as recited in claim 17 , wherein the associating comprises: determining that the broadcast television program has been watched for at least a minimum period of time; and associating a television program identifier that is associated with the broadcast television program with a VOD identifier that is associated with the video-on-demand.

20. The method as recited in claim 19 , wherein the minimum period of time comprises three minutes.

21. The method as recited in claim 17 , wherein the weight comprises a percentage of viewers who purchased the video-on-demand who also watched the broadcast television program.

22. The method as recited in claim 17 , wherein the weight is calculated based on association rules that are used to determine for multiple combinations of television programs that may be watched, a probability that a particular VOD will be purchased or watched.

23. The method as recited in claim 17 , wherein the recommending comprises determining that the particular viewer has not already purchased the video-on-demand.

24. The method as recited in claim 17 , wherein the recommending comprises determining that the video-on-demand is currently available for purchase by the particular viewer.

25. A computer-readable storage device having computer-readable instructions stored thereon that, when executed by a processor, implements the method as recited in claim 17 .

26. A method, comprising: transmitting television viewing history data from a client device to a server system; transmitting a video-on-demand recommendation request from the client device to the server system; and receiving, at the client device, a listing of available videos-on-demand that are recommended based on the television viewing history data compared to television viewing history data associated with other viewers who have purchased the recommended videos-on-demand, the association being based at least in part on assigning a weight to the television viewing history of the other viewers that is associated with the recommended videos-on-demand, wherein the weight describes a strength of association between the television viewing history of the other viewers and the recommended videos-on-demand and represents a conditional probability that a viewer will purchase a particular video-on-demand provided that the viewer has already watched a particular broadcast television program and a probability that any viewer will purchase the particular video-on-demand.

27. The method as recited in claim 26 , wherein the television viewing history data comprises: a client device identifier that uniquely identifies a client device through which broadcast television programs have been received; and a television program identifier that uniquely identifies a particular broadcast television program that has been watched.

28. A system, comprising: a network interface, at a headend, for receiving television data that identifies broadcast television programs that have been watched using one or more of a plurality of client devices; a network interface, at the headend, for receiving VOD data that identifies videos-on-demand that have been purchased through one or more of the plurality of client devices; a VOD recommendation engine, at the headend, configured to: associate a portion of the television data with a portion of the VOD data when the portion of the television data identifies a particular one of the broadcast television programs that was watched using a first client device and the portion of the VOD data identifies a particular one of the videos-on-demand that was purchased using the first client device, the associating including assigning a weight to the portion of the television data that is associated with the port of the VOD data, wherein the weight describes a strength of association between the portion of television data and the portion of VOD data and represents a conditional probability that a viewer will purchase the particular video-on-demand based on a factor that the viewer has already watched the particular broadcast television program and a probability that any viewer will purchase the particular video-on-demand; and generate a recommendation, to a second client device, that the particular video-on-demand be purchased based on received television data that indicates that the particular broadcast television program was watched using the second client device.

29. The system as recited in claim 28 , further comprising a TV data store for maintaining the television data.

30. The system as recited in claim 28 , further comprising a VOD data store for maintaining the VOD data.

31. The system as recited in claim 28 , wherein the first client device comprises a cable television set-top box.

32. The system as recited in claim 28 , wherein the first client device comprises a digital television recorder.

33. The system as recited in claim 28 , wherein the first client device comprises a personal computer.

34. The system as recited in claim 28 , wherein the system is implemented as a cable television system headend.

35. A system, comprising: means for maintaining television data, at a headend, that identifies broadcast television programs that are viewed using one or more of a plurality of client devices; means for maintaining video-on-demand data, at the headend, that identifies videos-on-demand that are purchased using one or more of the plurality of client devices; means for creating associations between broadcast television programs and available videos-on-demand, a particular one of the broadcast television programs being associated with a particular one of the videos-on-demand if the particular broadcast television program was watched and the particular video-on-demand was purchased using a particular one of the plurality of client devices, the means for creating associations including means for assigning a weight to the particular broadcast television program that is associated with the particular video-on-demand, wherein the weight describes a strength of association between the particular broadcast television program and the particular video-on-demand and represents a ratio that is defined as a conditional probability that a viewer will purchase the particular video-on-demand provided that the viewer has already watched the particular broadcast television program and a probability that any viewer will purchase the particular video-on-demand; and means for recommending the particular video-on-demand for purchase through a second one of the plurality of client devices based on a comparison of a portion of the television data that is associated with the second one of the plurality of client devices and the associations between broadcast television programs and available videos-on-demand.

36. The system as recited in claim 35 , wherein the means for recommending comprises a decision tree generation algorithm.

37. The system as recited in claim 35 , wherein the means for recommending comprises a data mining engine configured to apply association rules which generate probabilities that a particular VOD title will be purchased given that particular combinations of television programs have been viewed.

38. The system as recited in claim 35 , further comprising means for determining that the particular video-on-demand has not already been purchased using the second client device.

39. A computer-readable storage device comprising computer-readable instructions that, when executed, cause a computer system to: gather television data that identifies individual client devices and broadcast television programs that have been watched using the individual client devices; gather VOD data that identifies individual client devices and videos-on-demand that have been purchased using the individual client devices; and associate portions of the television data with portions of the VOD data, the associating including assigning a weight to the portions of the television data that are associated with the portions of the VOD data, the weight representing a likelihood that a particular viewer will purchase a particular one of the videos-on-demand based on a factor that the particular viewer has already watched a particular one of the broadcast television programs and a probability that any viewer will purchase the particular video-on-demand.

40. The one or more computer-readable storage device as recited in claim 39 , further comprising computer-readable instructions that, when executed, cause the computer system to apply a decision tree algorithm to the VOD data and the television data to create a decision tree for each of a plurality of VOD titles using the TV data to generate tree splitting criteria.

41. The one or more computer-readable storage device as recited in claim 40 , wherein: the decision tree for each of the plurality of VOD titles comprises a probabilistic classification tree; and the tree splitting criteria comprises a Bayesian score.

42. The one or more computer-readable storage device as recited in claim 39 , further comprising computer-readable instructions that, when executed, cause the computer system to apply association rules which generate probabilities that a particular VOD title will be purchased given that particular combinations of television programs have been viewed.

43. The one or more computer-readable storage device as recited in claim 39 , further comprising computer-readable instructions that, when executed, cause the computer system to recommend a particular video-on-demand to a particular viewer if the television data indicates that the particular viewer has watched a broadcast television program that is associated with the particular video-on-demand.

44. A computer-readable storage device comprising computer-readable instructions that, when executed, cause a computer system to: transmit to a server, television data and video-on-demand data that identifies broadcast television programs that have been watched and videos-on-demand that have been purchased using a particular client device; transmit to the server, a request for a video-on-demand recommendation from a second client device, the server associating the television data with the video-on-demand data; and receive, at the second client device, a recommendation for a particular one of the videos-on-demand, the recommendation having been generated based, at least in part, on an association between the television data and the video-on-demand data, the association including assigning a weight to the television data that is associated with the video-on-demand data, wherein the weight describes a strength of association between the television data and the VOD data and represents a conditional probability that a viewer will purchase a particular video-on-demand provided that the viewer has already watched a particular broadcast television program and a probability that any viewer will purchase the particular video-on-demand.